SDG corpora

SDG corpora is the independently-versioned home for the Signals Data Governance datasets used across the weathership stack — the “what” sibling of SDG strategy. Each commit is a coherent snapshot of one derivation:

OWL ⊨ SKOS ⊨ SHACL, then a relational projection.

The source of truth is a BFO 2020 / CCO-grounded OWL ontology, HermiT-certified. SKOS is not a second vocabulary: every published concept is a projection of a realized OWL class (labels, definitions, and examples live as SKOS annotation properties on that class). SHACL is the closed-world constraint view of the same axioms — sh:targetClass names a realized class; cardinality and filler restrictions become shape constraints. The relational schema (tables, keys, foreign keys, loadable SQL) is lowered from that certified OWL, not authored beside it. Generated textbook chapters and rows populate the projection; column labels come from the SKOS face.

Consumed via git submodule by three downstream projects:

• Ægir — training and evaluation of the hierarchical sequence model; owner of the derivation.
• Atelier — independent classification against the shared SKOS vocabulary.
• Signals — warehouse facts and Atlas tags against the populated relational footprint.

Tagged releases package the ontology, the SKOS vocabulary, the SHACL shapes, and the populated tables — without the per-column reference that would tell a downstream classifier which class minted each column. Atelier pins a release and classifies blind from values and vocabulary only; the held-back reference is the scoring key.

Released under the Apache 2.0 license.